1 / 15

Progress in the framework of the RESPITE project at DaimlerChrysler Research & Technology

Progress in the framework of the RESPITE project at DaimlerChrysler Research & Technology. Dr-Ing. Fritz Class and Joan Marí Martigny, Jan. 2002. Contents. DaimlerChrysler off-line demonstrator Block-diagram of our off-line demonstrator Next evaluation experiments using our demonstrator

taipa
Download Presentation

Progress in the framework of the RESPITE project at DaimlerChrysler Research & Technology

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Progress in the framework of the RESPITE project at DaimlerChrysler Research & Technology Dr-Ing. Fritz Class and Joan Marí Martigny, Jan. 2002

  2. Contents • DaimlerChrysler off-line demonstrator • Block-diagram of our off-line demonstrator • Next evaluation experiments using our demonstrator • On-going research in Discriminative Feature Extraction • TANDEM acoustic modelling • Linear-Discriminant-Analysis-based (LDA) front-end • Quadratic-Discriminant-Analysis-based (QDA) front-end • A two-layer perceptron to generate state-posteriors from QDA features (RBFs) • Results

  3. DC ASR system CTK/QUICKNET/MSTK DC off-line demonstrator: block-diagram

  4. DC off-line demonstrator: next steps • Evaluate results of IDIAP Multi-Stream toolkit on the AURORA 2000 database and compare them with those of SPRACHcore and CTK toolkits • Determine, given the results of the previous evaluation and system requirements, which is the desirable technique for our purposes • Using our own in-car american english database compare our baseline system with the selected optimum technique

  5. Contents • DaimlerChrysler off-line demonstrator • Block-diagram of our off-line demonstrator • Next evaluation experiments using our demonstrator • On-going research in Discriminative Feature Extraction • TANDEM acoustic modelling • Linear-Discriminant-Analysis-based (LDA) front-end • Quadratic-Discriminant-Analysis-based (QDA) front-end • A two-layer perceptron to generate state-posteriors from QDA features (RBFs) • Results and Conclusions

  6. Non-linear transform of the feature space Neural Net training Discriminative Feature Extraction:TANDEM training

  7. Linear transform to reduce dimensionality Supervised Clustering Discriminative Feature Extraction:LDA training

  8. TANDEM training LDA training Bayes rule Discriminative Feature Extraction: LDA

  9. TANDEM features are obtained from log-posteriors Applying Bayes rule as in the previous slide A quadratic equation is obtained Discriminative Feature Extraction: QDA TANDEM can be interpreted as a kind of non-linear feature extraction • Key questions at this point are: • Is one gaussian per cluster enough ? • How many classes should be used ? • Is the gaussianity assumption always a good one?

  10. Returning back to the Bayes rule An RBF is thus obtained We could express it as: Where f is the softmax function and N is the gaussian pdf Discriminative Feature Extraction: RBFs A compromise between connectionist and parametric modelling are RBFs

  11. Discriminative Feature Extraction: results Recognition results on AURORA 2000

  12. Discriminative Feature Extraction: results Recognition results on AURORA 2000

  13. Discriminative Feature Extraction: Results Recognition results on AURORA 2000

  14. Reduction of the dimensionality of the Neural Net Discriminative Feature Extraction: results

  15. Discriminative Feature Extraction: conclusions • TANDEM acoustic modelling can be performed with discriminant parametric models too (QDA) • As a compromise between connectionist and parametric modelling RBFs can be used for TANDEM • Concatenation of LDA-PLP and LDA-MSG features results in an slight improvement to our baseline LDA system • Word-based Hybrid ANN/HMMs are the best performing

More Related